Since the start of the year, the US software complex has de-rated sharply. The S&P North American technology software proxy (IGV) was down roughly 25% at the February lows before rebounding, but the broader message has been clear: selling has been broad-based across Software as a Service. Software as a Service (Saas) is the licensing of software over the internet instead of installing and maintaining it on your own servers.

The market has been spooked by what AI implies for software business models. The core fear is that SaaS pricing, particularly per-seat and per-licence models, becomes less defensible if AI materially lifts productivity and customers need fewer users to achieve the same output. At the same time, AI is reducing the cost and complexity of building software. As development becomes easier and less resource-intensive, the historical IP moat that protected many incumbents looks less durable, putting long-term pricing power under pressure.

For the perspective of an equity analyst, this ultimately a terminal value debate. For fast-growing tech companies that traded on high multiples, a large share of valuation sits in what investors assume the business can earn well beyond the next few years. When the market starts questioning long-run pricing power, market position and moat durability, terminal assumptions compress and multiples contract, even if near-term consensus cash flows haven’t changed.

ADOBE (ADBE)
Nasdaq listed Adobe Inc. illustrates that dynamic clearly. The stock is down about 59% from around $635 to the mid-$200s, yet over that period the fundamentals have gone the other way. FY25 revenue increased to $23.77bn from $21.51bn in FY24 (up about $2.26bn, or 11%), net income increased to $7.13bn from $5.56bn (up about $1.57bn, or 28%), and free cash flow (operating cash flow less capex) increased to roughly $9.85bn from $7.87bn (up about $1.98bn, or 25%).

More recently, the disconnect has become sharper. Over the past six months, the stock is down 21.5% even as earnings revisions have skewed positive, with 29 analysts lifting their EPS estimates versus nine taking them down.

In other words, near-term expectations are improving, and results appear to be tracking ahead of prior assumptions, yet the share price has continued to drift lower. That is usually a signal the market is not debating the next year — it’s debating the durability of the long-run cash flows and repricing the terminal value it is willing to capitalise.

On Adobe specifically, the concern is that its historical profit engine is vulnerable to “AI unbundling.” Adobe’s model has long been anchored in high-frequency professional workflows (Creative Cloud), entrenched file standards, and organisational switching costs built through training, templates, plug-ins, and standardised team processes. That created pricing power because replacing Adobe wasn’t just switching software — it meant retraining people and rebuilding workflows.

Generative AI threatens Adobe’s moat by shifting value from mastering Adobe’s tools to simply getting an outcome. As creation becomes prompt-led and automated, the learned-skill switching cost compresses and the interface matters less—creative work can be done through assistants, AI-native apps, or tools embedded outside Adobe.

AI may also commoditise core creative capabilities. Many features that once required differentiated engineering may become more easily available as making credible alternatives becomes faster and cheaper to build. That lowers barriers to entry, speeds feature parity, and expands the set of good-enough, low-cost, or free options.

The risk is not near-term collapse; it’s long-run erosion in pricing power and premium economics. If creative capability becomes abundant and interchangeable, Adobe’s differentiation narrows and the value it can provide diminishes. Halting its ability to sustain premium pricing a decade from now.

Market reaction to AI disruption
AI defensiveness isn’t uniform. Different software categories have quite different exposure depending on where they sit in a customer’s workflow. Many incumbents argue their moat isn’t the software IP— but the combination of rich proprietary data, embedded compliance and regulatory integrations, long-lived customer relationships, and being deeply embedded as a mission-critical system of record.

In that framing, the debate shifts away from horizontal feature sets (where AI can compress differentiation quickly) toward vertical platforms where accuracy, trust, connectivity, and workflow ownership matter more than the UI.

Wise Tech is a good example of how a vertical operator positions this. Management has been clear that CargoWise is not “software sitting on top of workflows” — it is the execution layer for government-regulated customs, compliance, transport, and documentation across many jurisdictions, with permissioned data flows embedded inside operating systems.

The argument is that the hard part isn’t writing code; it’s the regulated connectivity, the legal business rules, the domain expertise, and operational correctness at scale. In that worldview, AI is more likely to be absorbed as a productivity layer inside the platform than something that fully substitutes it.

The counter-argument is that the incentive to build in-house is rising. As software development becomes more efficient, the business case for internalising mission-critical workflows becomes more compelling: capture the ownership economics, reduce vendor dependence, and potentially build a competitive edge.

But it’s a double-edged sword. The downside is spending millions and years building a worse product, then reverting back to the vendor solution that reflects decades of iteration, deep expertise, and cumulative investment.

Large, sophisticated operators with the scale and resources like DSV (the world’s largest freight forwarder) have indicated that they may bring these capabilities in house, but it is not without risk, and the transition process is likely to take years.

AI: friend or foe?
Ultimately, to have an investment view on any SaaS company today, you have to take a view on what AI does to its business model. For some companies, AI will be a tailwind, and management teams are already pointing to large productivity gains — including talk of headcount reductions of up to 50% in some areas as AI automates parts of the workflow (like we have seen for Wise Tech and Block). If revenue models remain intact and topline growth continues, that operating leverage can translate into faster earnings growth and stronger cash generation, strengthening rather than weakening the case for premium multiples.

The market used to reward these companies with lofty premiums because investors believed they could compound free cash flow far into the future with limited disruption. That confidence is now being tested. Execution risk has been replaced by business-model risk. Growth assumptions embedded in many valuations are less demanding than they were at the peak, but the range of outcomes has widened. This will prove to be either a generational buying opportunity or a period of real value destruction.

James Cameron is a Junior Equity Analyst at Octagon Asset Management.

This article has been prepared in good faith based on information obtained from sources believed to be reliable and accurate. This article does not contain financial advice. Some of the Octagon portfolios may own securities issued by companies mentioned in this article.

Octagon Asset Management is the investment manager for Octagon Investment Funds and the Summer KiwiSaver scheme.